Using Gas-Phase Modeling To Track The Fate Of Pure Oxygen At A Large HPO Facility

Introduction
Biological process models for water resource recovery facilities are now standard tools in the industry. Commercially available modeling software at present incorporates aeration calculations into the model to simulate oxygen transfer. However, these simulators offer limited support for simulating oxygen transfer for high-purity oxygen activated sludge (HPOAS) facilities. Often, a design engineer will have to change the composition of the supply gas manually, zone-by-zone, through the treatment basin to reflect the gradual depletion of oxygen in the supply along the length of the treatment basin. Some more recent simulators do allow oxygen gas to transfer zone by zone through the process while reflecting gas transfer via surface aerators, which are significant improvements. However, this model still assumes that 100% of the fed oxygen is either dissolved or vented at the downstream end of treatment basin, and this may not reflect full-scale systems. Most HPOAS plants in the US are housed in structures that are at least 30 years old and perfectly sealed enclosed gas headspaces for these facilities is not a reasonable assumption. The objectives of this paper are to demonstrate the importance of utilizing gas-phase modeling when simulating an HPOAS plant and to demonstrate the benefits of gas-phase modeling with a case-study of a large facility in Tennessee. <
b>Background
The Moccasin Bend Wastewater Treatment Plant (MBWWTP) is located in Chattanooga, TN. The plant treats influent from a combined sewer and is currently capable of treating up to 140 MGD through the primary and secondary treatment systems. Secondary treatment is accomplished through a four-train UNOX HPOAS activated sludge facility, each train with eight cells in series. A biological process model was built in SUMO based on the existing unit processes at the MBWWTP to simulate process performance. The model included the primary clarifiers, equalization basin, UNOX bioreactors, secondary clarifiers, and the solids train. Critically, the model utilized Jacobs' proprietary model units for gas-phase modeling through the UNOX basins. These units allowed the model to replicate the feed pure oxygen from the oxygen plant and the vented oxygen at the downstream end of the UNOX basins. A process flow diagram of the SUMO model is depicted in Figure A.
Methodology
To overcome these weaknesses, a new enclosed CSTR module was developed using the SUMO simulation platform, by Dynamita, that includes an enclosed headspace and separate inputs and outputs for gas and liquid. The enclosed CSTR uses SUMO's embedded aeration calculations and is compared to a typical HPOAS facility schematic in Figure B. These inputs and outputs allow for a design engineer to accurately model the fate of the oxygen gas in an operating HPOAS facility, accounting for real world conditions such as leaks, variation in surface aerators efficiency, and the realities of vent purity control. The MBWWTP process model was calibrated with historical diurnal plant data provided for the month of February 2021 and validated with data from November 2021, following the IWA Good Modeling Practice Guidelines.
Findings
Liquid Phase Model Calibration Model calibration was achieved in both steady-state and with dynamic daily data. Average influent conditions used in the model calibration are shown in Table A. The model calibration process went through several steps, the first of which was approximating the actual efficiency of the installed surface aerators to obtain a rough match between recorded and simulated DO concentrations. Kinetic adjustments were also made to account for nitrification inhibition. Calibration and validation results are shown in Tables B and C, respectively.
Gas Phase Model Calibration For the gas phase, a comparison of the simulated oxygen vent purity data from the downstream end of the UNOX basins with the measured vent purity showed substantial discrepancies. Initially, the SUMO model was set up so that all the oxygen fed into the system was either transferred or vented, which led to a predicted vent purity well over 80%. Vent purities for UNOX 1 and 2 trend around 70% while UNOX trains 3 and 4 were substantially lower. Raw data from the vent purity meters for February 2021 is provided in Figure C. UNOX 3 vent purity data read a constant 21% nearly every day, raising suspicions that the meter was reading atmospheric oxygen conditions. Subsequent investigation showed that the UNOX 3 vent pipe had been clogged for a significant time, implying that the vent gas was using another exit. Further investigation with a lower explosive limit (LEL) sensor identified roof leaks on the seams between UNOX reactor zones. The LEL produced an out-of-range reading when placed on these seams, implying that the open air in that location had an O2 content above 30% (Figure D).
The roof leaks were incorporated into the model to match the lower vent purity data readings. Simulating UNOX 3 required assuming 100% of the exhaust gas was leaked through roof seams and UNOX 4 would have required simulating leaks above 90% of non-transferred oxygen. These values are extreme, and as a result, the model was calibrated against the average vent purity from trains 1 and 2, as those trains appeared to have a significant mass of oxygen reaching the vent. The data for UNOX trains 1 and 2 were still substantially lower than a leak free model would predict so three different gas leaks were added to simulate the leaked gas from UNOX 1 and 2. The Sumo model with the leaks is depicted in Figure E and the mass of oxygen leaking through the roof in the simulation is shown in Table D. The model predicted that approximately 52% of the oxygen fed to the UNOX is lost to the system through roof leaks in the two best performing trains in the plant. The vent purity output from the model is compared to the averaged data from the two trains in Figure F.
Status
The modeling study has been completed with a short-term recommendation that the City seal the roof seams to mitigate the leaks.
Significance
- It is important to understand the mechanics of unit processes included in a process model. Understanding the limitations of existing HPOAS simulators lead to a more realistic unit process model for handling HPOAS facilities. - Models that do not simulate cell to cell gas transfer do not provide a realistic simulation of an HPOAS facility. Similarly, models that assume that all gases in an enclosed headspace are transferred to the subsequent cell are not realistic simulations of aged facilities. - Gas leaks at pure oxygen plants can be very large. The scale of the leaks at MBWWTP are now estimated to be an order of magnitude greater than previously assumed, representing a substantial energy loss.